Quantum computing will make supercomputers hilariously slow by comparison. How will this impact AI's pace of change to our society?
AI has been trained using — and was built on — classic computing power
AI is improving faster and faster — an accelerating pace of change
Now, trillions of dollars is being poured into AI to scale it up even further, and faster
When we are able to utilize quantum computing for AI, the pace of change explodes even faster
What does recursive self-improvement look like in the quantum age?
AI has been trained using — and was built on — classic computing power
AI is improving faster and faster — an accelerating pace of change
Now, trillions of dollars is being poured into AI to scale it up even further, and faster
When we are able to utilize quantum computing for AI, the pace of change explodes even faster
What does recursive self-improvement look like in the quantum age?
One of the goals with AI models is to get to the point of recursive self-improvement, which means the AI can — with autonomy — work in an endless loop to learn more, become smarter, and generally improve.
Today, that would happen on so-called "classic" computing, which are just the normal processors at work in your phones, TVs, data centers, and supercomputers. These are definitely fast and keep getting faster, and there's specialized versions that make some tasks massively faster.
Without getting too deep in the mechanics, quantum computing is basically a computing holy grail. However fast we can make classic computing chips, it pales in comparison to the speed of quantum computing.
Compared to quantum computers, computers today would seem like the room-sized computers of the 1950s. It's a fighter jet vs. a hot air balloon. It's a rifle bullet vs. a paper airplane.
Now think about the progress we've made with AI, and how fast — a few years ago AI was hilariously terrible with Will Smith eating spaghetti. Now it's lifelike and can fool you. Understand this was a few years using classical computing. Now imagine quantum computing becomes widely available for use in AI applications, specifically to recursively self-improve. What does the improvement look like in 3 years? Can we conceive of it as a species?